forked from plusjade/jekyll-bootstrap
-
Notifications
You must be signed in to change notification settings - Fork 0
/
rbm.py
366 lines (327 loc) · 14 KB
/
rbm.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
# -*- coding: utf-8 -*-
"""
Created on Tue Sep 18 14:52:30 2012
@author: zhengxin
"""
#==============================================================================
# This file contains three(four) kinds of RBMs: Gaussian-Bernoulli RBM,
# Bernoulli-Bernoulli RBM, Discriminative RBM
#==============================================================================
from options import options
import numpy as np
import cudamat as cm
from softmax import softmax
import scipy.io
import os
class GaussianRBM(object):
def __init__(self, datafolder=None, num_hid=None, options=None):
if datafolder == None:
return
self.datafolder = datafolder
self.datalist = os.listdir(datafolder)
self.num_batchdata = len(self.datalist)
mdict = scipy.io.loadmat(os.path.join(datafolder, self.datalist[0]))
tempdata = mdict['data']
self.options = options
self.num_vis = tempdata.shape[0]
self.num_hid = num_hid
# print self.num_vis
# print self.num_hid
self.num_batches = tempdata.shape[1]/self.options.batchsize
self.batch_size = self.options.batchsize
self.doPCD = False
self.cdstep = 1
# initialize weights
self.W = cm.CUDAMatrix(0.01 * np.random.randn(self.num_vis, self.num_hid))
self.vb = cm.CUDAMatrix(np.zeros((self.num_vis,1)))# for gaussian rbm, v_bias we mean the mean of visible layer
self.hb = cm.CUDAMatrix(np.zeros((self.num_hid,1)))
# initialize weights updates
self.dW = cm.CUDAMatrix(np.zeros((self.num_vis, self.num_hid)))
self.dvb = cm.CUDAMatrix(np.zeros((self.num_vis, 1)))
self.dhb = cm.CUDAMatrix(np.zeros((self.num_hid, 1)))
self.W_inc = cm.CUDAMatrix(np.zeros((self.num_vis, self.num_hid)))
self.vb_inc = cm.CUDAMatrix(np.zeros((self.num_vis,1)))
self.hb_inc = cm.CUDAMatrix(np.zeros((self.num_hid,1)))
# initialize temporary storage
self.v = cm.empty((self.num_vis, self.batch_size))# a batch of data
self.vm = cm.empty((self.num_vis, self.batch_size))# temp storage of data-vb
self.h = cm.empty((self.num_hid, self.batch_size))
self.r = cm.empty((self.num_hid, self.batch_size))# store random number in positive phase
self.r2 = cm.empty((self.num_vis, self.batch_size))# store random number in negative phase
def getCurrentBatch(self,mdict,batch):
# get current batch
batchdata = mdict['data'][:,batch*self.batch_size:(batch+1)*self.batch_size]
self.v = cm.CUDAMatrix(batchdata)
self.v_true = cm.CUDAMatrix(batchdata)
def applyMomentum(self):
# apply momentum
# maybe we can change it while proccessing
self.dW.mult(0)
self.dvb.mult(0)
self.dhb.mult(0)
self.W_inc.mult(self.options.momentum)
self.vb_inc.mult(self.options.momentum)
self.hb_inc.mult(self.options.momentum)
def hidActProb(self,vis, target):
# positive phase
# print self.W.shape
# print vis.shape
# print target.shape
cm.dot(self.W.T, vis, target = target)
target.add_col_vec(self.hb)
target.apply_sigmoid()
def visActProb(self):
# negative phase
cm.dot(self.W, self.h, target = self.v)
self.v.add_col_vec(self.vb)#now v = Wh + c
def CDstats(self, vis, hid, posphase=True):
multiplier = 1.0 if posphase else -1.0
self.dhb.add_sums(hid, 1, mult=multiplier)
if posphase:
#print 'posphase'
self.dW.add_dot(vis, hid.T)
self.vm.assign(vis)
self.vb.mult(-1)
self.vm.add_col_vec(self.vb)
self.vb.mult(-1)
self.dvb.add_sums(self.vm, 1, mult=multiplier)
else:
#print 'negphase'
self.dW.subtract_dot(vis,hid.T)
self.vm.assign(vis)
self.vb.mult(-1)
self.vm.add_col_vec(self.vb)
self.vb.mult(-1)
self.dvb.add_sums(self.vm, 1, mult=multiplier)
def sampleHid(self,r,target):
# sample hiddens
r.fill_with_rand()
r.less_than(target, target = target)
def sampleVis(self):
self.r2.fill_with_randn()
self.v.add(self.r2)
def CDn(self):
n = self.cdstep
self.hidActProb(self.v, self.h)
self.CDstats(self.v, self.h)
for i in range(n):
self.sampleHid(self.r,self.h)
self.visActProb()
self.sampleVis()
self.hidActProb(self.v, self.h)
self.CDstats(self.v, self.h, False)
def doOneStep(self):
if self.doPCD:
self.PCD()
else:
self.CDn()
self.updateWeights()
def updateWeights(self):
self.W_inc.add_mult(self.dW, self.options.eta/self.batch_size)
self.vb_inc.add_mult(self.dvb, self.options.eta/self.batch_size)
self.hb_inc.add_mult(self.dhb, self.options.eta/self.batch_size)
# update weights
self.W.add(self.W_inc)
self.vb.add(self.vb_inc)
self.hb.add(self.hb_inc)
def getReconErr(self):
self.v.subtract(self.v_true)
return self.v.euclid_norm()**2
def train(self):
for epoch in range(self.options.maxepoch):
err = []
for batchdata in range(self.num_batchdata):
mdict = scipy.io.loadmat(os.path.join(self.datafolder, self.datalist[batchdata]))
#data = mdict['data']
for batch in range(self.num_batches):
self.getCurrentBatch(mdict,batch)
self.doOneStep()
self.applyMomentum()
err.append(self.getReconErr()/(self.num_vis*self.batch_size))
print "Epoch " + str(epoch + 1)+" "+"Mean squared error: " + str(np.mean(err))
def getDataUpwards(self,loadfolder,savefolder):
# push data of visible layer upwards to form a set of new data
# because of memory issues, we have to each batch data to disc and read and combine them later
# batch mode receive data from cpu and return a matrix on cpu
datalist = os.listdir(loadfolder)
batchsize = 4096
n = 0
for dataname in datalist:
name = os.path.join(loadfolder,dataname)
mdict = scipy.io.loadmat(name)
data = mdict['data']
labels = mdict['label']
# print labels.shape
numbatch = data.shape[1]/batchsize
for batch in range(numbatch):
#print 'batch %d/%d'%(n, numbatch*len(datalist))
batchdata = data[:,batch*batchsize:(batch+1)*batchsize]
batchlabels = labels[batch*batchsize:(batch+1)*batchsize]
temp = cm.empty((self.num_hid,batchdata.shape[1]))
vis = cm.CUDAMatrix(batchdata)
self.hidActProb(vis, temp)
temp.copy_to_host()
#topdata[:,batch*batchsize:(batch+1)*batchsize] = temp.numpy_array
mdict = {}
mdict['data'] = temp.numpy_array
mdict['label'] = batchlabels
scipy.io.savemat('%s/%d.mat'%(savefolder,n),mdict)
n = n+1
def getTestDataUpwards(self,data):
batchsize = 4096
numbatch = data.shape[1]/batchsize
topdata = np.zeros((self.num_hid,data.shape[1]))
for batch in range(numbatch):
batchdata = data[:,batch*batchsize:(batch+1)*batchsize]
temp = cm.empty((self.num_hid,batchdata.shape[1]))
vis = cm.CUDAMatrix(batchdata)
self.hidActProb(vis, temp)
temp.copy_to_host()
topdata[:,batch*batchsize:(batch+1)*batchsize] = temp.numpy_array
return topdata
def save(self,filename):
self.W.copy_to_host()
self.vb.copy_to_host()
self.hb.copy_to_host()
mdict = {}
mdict['type']='gauss'
mdict['W']=self.W.numpy_array
mdict['vb']=self.vb.numpy_array
mdict['hb']=self.hb.numpy_array
scipy.io.savemat(filename,mdict)
def load(self, filename):
mdict = scipy.io.loadmat(filename)
self.W = cm.CUDAMatrix(mdict['W'])
self.vb = cm.CUDAMatrix(mdict['vb'])
self.hb = cm.CUDAMatrix(mdict['hb'])
(self.num_vis, self.num_hid) = self.W.shape
class BinaryRBM(GaussianRBM):
def visActProb(self):
GaussianRBM.visActProb(self)
self.v.apply_sigmoid()
def CDstats(self, vis, hid, posphase=True):
multiplier = 1.0 if posphase else -1.0
self.dhb.add_sums(hid, 1, mult=multiplier)
self.dvb.add_sums(vis, 1, mult=multiplier)
if posphase:
self.dW.add_dot(vis, hid.T)
else:
self.dW.subtract_dot(vis,hid.T)
def sampleVis(self):
# sample hiddens
self.r2.fill_with_rand()
self.r2.less_than(self.v, target = self.v)# now h = phstates
class SoftmaxRBM(BinaryRBM):
def hidActProb(self,vis, target):
cm.dot(self.W.T, vis, target = target)
target.add_col_vec(self.hb)
softmax(target)
class DiscriminativeRBM(GaussianRBM):
def __init__(self,datafolder=None,labels=None,numhid=None,options=None):
# the labels here is just used for calculating number of labels, no other practical use.
super(DiscriminativeRBM,self).__init__(datafolder,numhid,options)
self.labels = labels
self.num_class = self.getClassNum(labels)
#self.targets = self.getTargets()
if datafolder == None:
return
self.cW = cm.CUDAMatrix(0.01 * np.random.randn(self.num_class,self.num_hid))
self.cb = cm.CUDAMatrix(np.zeros((self.num_class,1)))
self.dcW = cm.CUDAMatrix(np.zeros((self.num_class,self.num_hid)))
self.dcb = cm.CUDAMatrix(np.zeros((self.num_class,1)))
self.cW_inc = cm.CUDAMatrix(np.zeros((self.num_class,self.num_hid)))
self.cb_inc = cm.CUDAMatrix(np.zeros((self.num_class,1)))
self.c = cm.empty((self.num_class,self.batch_size))
def getClassNum(self,labels):
self.labellist = np.unique(labels)
return len(self.labellist)
def getTargets(self,labels):
# create targets
targets = np.zeros((self.num_class,len(labels)))
#print targets.shape
for i in range(self.num_class):
for j in range(len(labels)):
if labels[j] == self.labellist[i]:
targets[i,j] = True
return targets
def applyMomentum(self):
super(DiscriminativeRBM,self).applyMomentum()
self.dcW.mult(0)
self.dcb.mult(0)
self.cW_inc.mult(self.options.momentum)
self.cb_inc.mult(self.options.momentum)
def hidActProb(self,vis, target):
# positive phase
cm.dot(self.W.T, vis, target = target)
target.add_dot(self.cW.T, self.c)
target.add_col_vec(self.hb)
target.apply_sigmoid()
def getCurrentBatch(self,mdict,batch):
super(DiscriminativeRBM,self).getCurrentBatch(mdict,batch)
#print mdict['label'].shape
batchlabels = mdict['label'][batch*self.batch_size:(batch+1)*self.batch_size]
batchtargets = self.getTargets(batchlabels)
self.c = cm.CUDAMatrix(batchtargets)
def CDstats(self, vis, hid, posphase=True):
multiplier = 1.0 if posphase else -1.0
self.dhb.add_sums(hid, 1, mult=multiplier)
self.dvb.add_sums(vis, 1, mult=multiplier)
if posphase:
self.dW.add_dot(vis, hid.T)
self.dcb.add_sums(self.c, 1, mult=1.0)
self.dcW.add_dot(self.c, hid.T)
else:
self.dW.subtract_dot(vis,hid.T)
self.dcb.add_sums(self.c, 1, mult=-1.0)
self.dcW.subtract_dot(self.c,hid.T)
def visActProb(self):
# negative phase
super(DiscriminativeRBM,self).visActProb()
self.v.apply_sigmoid()
cm.dot(self.cW, self.h, target = self.c)
self.c.add_col_vec(self.cb)
softmax(self.c)
def updateWeights(self):
super(DiscriminativeRBM,self).updateWeights()
self.cW_inc.add_mult(self.dcW, self.options.eta/self.batch_size)
self.cb_inc.add_mult(self.dcb, self.options.eta/self.batch_size)
self.cW.add(self.cW_inc)
self.cb.add(self.cb_inc)
def save(self,filename):
self.W.copy_to_host()
self.vb.copy_to_host()
self.hb.copy_to_host()
self.cb.copy_to_host()
self.cW.copy_to_host()
mdict = {}
mdict['type']='discriminative'
mdict['W']=self.W.numpy_array
mdict['vb']=self.vb.numpy_array
mdict['hb']=self.hb.numpy_array
mdict['cb']=self.cb.numpy_array
mdict['cW']=self.cW.numpy_array
scipy.io.savemat(filename,mdict)
def load(self, filename):
mdict = scipy.io.loadmat(filename)
self.W = cm.CUDAMatrix(mdict['W'])
self.vb = cm.CUDAMatrix(mdict['vb'])
self.hb = cm.CUDAMatrix(mdict['hb'])
self.cb = cm.CUDAMatrix(mdict['cb'])
self.cW = cm.CUDAMatrix(mdict['cW'])
# def train(self):
# for epoch in range(self.options.maxepoch):
# #err = 0
# err = []
# for batchdata in range(self.num_batchdata):
# #print 'batchdata'+str(batchdata)
# #print self.datalist[batchdata]
# mdict = scipy.io.loadmat(os.path.join(self.datafolder, self.datalist[batchdata]))
# data = mdict['data']
# for batch in range(self.num_batches):
# #print 'batch'+str(batch)
# self.getCurrentBatch(data,batch)
# self.doOneStep()
# self.applyMomentum()
# err.append(self.getReconErr()/(self.num_vis*self.batch_size))
# print "Epoch " + str(epoch + 1)+" "+"Mean squared error: " + str(np.mean(err))
#